The Economics of Parenting

EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

Annual Review of Economics The Economics of Parenting

Matthias Doepke,1 Giuseppe Sorrenti,2 and Fabrizio Zilibotti3 1Department of Economics, Northwestern University, Evanston, Illinois 60208, USA; email: [email protected] 2Department of Economics and Jacobs Center for Productive Youth Development, University of Zurich, Zurich 8001, Switzerland; email: [email protected] 3Department of Economics, Yale University, New Haven, Connecticut 06511, USA; email: [email protected]

Annu. Rev. Econ. 2019. 11:55–84 Keywords First published as a Review in Advance on parenting, parenting style, skill acquisition, peer effects, altruism, May 13, 2019 paternalism The Annual Review of Economics is online at economics.annualreviews.org Abstract Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org https://doi.org/10.1146/annurev-economics- Access provided by Northwestern University on 10/02/19. For personal use only. Parenting decisions are among the most consequential choices that peo- 080218-030156 ple make throughout their lives. Starting with the work of pioneers such Copyright © 2019 by Annual Reviews. as Gary Becker, economists have used the tool set of their discipline to un- All rights reserved derstand what parents do and how parents’ actions affect their children. In JEL codes: J13, J24, R20 recent years, the literature on parenting within economics has increasingly leveraged findings and concepts from related disciplines that also dealwith parent–child interactions. For example, economists have developed models to understand the choice among various parenting styles that were first explored in the developmental psychology literature and have estimated detailed empirical models of children’s accumulation of cognitive and noncognitive skills in response to parental and other inputs. In this review, we survey the economic literature on parenting and point out promising directions for future research.

55 EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

1. INTRODUCTION Among the distinct characteristics of humans is extraordinarily slow development. We are born helpless and then go through a prolonged phase of childhood and adolescence before reaching full maturity decades into our lives. As a consequence, young humans depend on their parents. Child rearing or parenting refers to everything that parents do to support the development of their children, from basics such as providing food and shelter to guiding their emotional and intellectual development. In the past parenting was mostly studied in other social sciences such as sociology and developmental psychology; however, in recent years, a large literature on parenting has developed within economics. In part, this reflects the growing reach of economics, which in the case of thestudyof parenting goes back to Gary Becker’s work on the family (see Becker 1981). However, an equally important factor is the fact that recent research shows that parenting decisions have profound economic implications. In modern economies, approximately two-thirds of total income goes to workers as wages, and most of this income reflects the return to accumulated human capital rather than raw labor. A key factor in the accumulation of human capital, in turn, is parenting decisions that start in the very first years of life and continue through a child’s development to adulthood. Economists who want to study the role of human capital in economic development, determinants of social mobility, and the drivers of long-run inequality must take parenting decisions into account. Choices that matter for parenting start long before a child is born. In the marriage market, for example, one criterion in choosing a partner is what she or he has to offer in terms of expected future investments in children. Several papers on the economics of marriage analyze how future parenting plans affect premarital investments and the matching patterns in the marriage market (see, for example, Chiappori et al. 2017). In turn, these choices influence the parenting choices that take place after a relationship is formed. Even more directly, the choice of how to raise each individual child is closely linked to the choice of how many children to have in the first place. One of the central concepts in the economics of fertility is Becker’s quantity–quality trade-off, i.e., the notion that a larger number of children (quantity) tends to lower investment in each individual child (quality).1 This trade-off is inevitable, particularly in regard to parents’ time: For parents with many children, it is simply not possible to spend hours of one-on-one time each day with each individual child, whereas parents of only children have more flexibility. Fertility and other parenting decisions are therefore closely linked. Nevertheless, in this review, we focus more narrowly on child-rearing decisions that take place after parents have formed a relationship and have had a child. In part, this is because the study of the Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org economics of marriage and fertility has a longer history,and many surveys are available.2 Moreover, Access provided by Northwestern University on 10/02/19. For personal use only. to date, the literatures on marriage, fertility,and narrowly defined parenting have mostly developed in isolation, with a lack of research that considers, say, a joint decision involving how many children to have and with which parenting style to raise them. We believe considering interactions among different choices that affect child rearing to be an important direction for future research, and we discuss some specific issues (such as the role of single parenting) at the end of the review.

1Becker (1960) provides the initial statement of the quantity–quality tradeoff, and Doepke (2015) provides an overview of the role of this concept in the economics of fertility. 2For an overview of the economics of marriage and the family and some recent directions in this literature, the reader is referred to Browning et al. (2014), Lundberg & Pollak (2007), and Lundberg et al. (2016). Surveys with a focus on the macroeconomic implications of family decisions are provided by Doepke & Tertilt (2016) and Greenwood et al. (2017). For the economics of fertility, the reader is referred to, for example, Hotz et al. (1997) and Doepke & Zilibotti (2019, chapter 7).

56 Doepke • Sorrenti • Zilibotti EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

We develop our discussion of the economic literature on post-birth parenting around three main themes. First, we focus on the choice of a parenting style. The term parenting style stems from developmental psychology and refers to the broad strategies that parents follow in raising their children, such as an obedience-based authoritarian style versus permissive parenting that grants more independence to children. We describe how the economics of parenting can be used to derive predictions on how economic conditions affect the choice of parenting style, and we point to empirical evidence suggesting that variation in parenting practices over time and across space lines up well with these predictions. Second, we focus on technology, and in particular on the production function allowing children to acquire and accumulate skills. Some of the issues discussed in the underlying literature are the relative importance of parental and other inputs, the effect of early versus late inputs as children develop, and the complementarity among inputs over time and across different types of inputs. Third, we focus on the effects of the environment. It has long been recognized that peer effects count for a lot in child development, especially during adolescence. This suggests that parents’ impact on their children may work in large part through their influence on their children’s environment, rather than through their direct interactions with them. For example, parents choose the neighborhood in which a family lives and the schools that their children attend. We discuss how these choices can be framed as an integral part of the economics of parenting. We organize our discussion by presenting a general model that captures the main aspects of the economics of parenting. We model the decision problem of parents who care about their own utility and the welfare of their children. Following the work of Doepke & Zilibotti (2017), the concern for children allows for both altruism (parents care about their children’s utility) and paternalism (parents care about their children’s actions in ways that potentially conflict with the children’s own preferences). Parents face a set of constraints that include not only standard budget and time constraints, but also technological constraints. For example, there is a production function that specifies how children’s skills and attitudes evolve as a function of endowments and parental inputs. The framework also allows for environmental influences outside the immediate family, such as peer effects in schools and neighborhood effects, to affect child development. Using this general framework, we discuss the economic literature on the three main issues outlined above, and we relate theoretical results to empirical evidence. Much of the literature on the economics of parenting is recent; the research frontier is constantly evolving, and many new questions raised by current research are yet to be addressed. We therefore conclude by outlining a few areas for future research on the economics of parenting that we consider particularly promising. In Section 2, we present our economic model of parenting. In Section 3, we use this model to

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org discuss the economics of choosing a parenting style and compare the predictions of the theory to Access provided by Northwestern University on 10/02/19. For personal use only. data. In Section 4, we relate the recent literature on children’s skill acquisition to our model of parenting. Neighborhood and peer effects are discussed in Section 5. Conclusions and directions for future research are discussed in Section 6.

2. AN ECONOMIC MODEL OF PARENTING In this section, we outline a general economic model of parenting that we use below to discuss parenting styles, skill formation, and neighborhood and peer effects. What we mean by economic model is that we conceive of parenting decisions in the same way that economists conceive of any other decision: Parents pursue particular objectives that are summarized by a utility function; they act rationally and purposefully in pursuing these objectives; and they are subject to various constraints such as limits to their financial resources, their knowledge, their time, and the underlying technology of child development. Given that much of parenting is about parent–child interactions, the model also allows for children to have a say and take actions on their own. The main

www.annualreviews.org • The Economics of Parenting 57 EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

elements of the model are a combination of work on parenting style by Doepke & Zilibotti (2017) with the notion of a multistage production function for skills, as in James Heckman’s recent work with different coauthors (e.g., Cunha & Heckman 2007, Cunha et al. 2010). In the model, we limit attention to a single parent and a single child. We start with the parent’s objectives. The parent derives utility during two periods of adulthood, and also cares about the child. The value function that the parent seeks to maximize is given by = | + | + − γ v + γ v . V U1(C1, L1 A) U2(C2, L2 A) Z [(1 ) ˜ ] 1.

We use the convention that upper-case variables refer to the parent, and lower-case variables to

the child. Accordingly, Ct and Lt are the parent’s consumption and leisure in period t,andA is a vector of preference parameters for the parent. The overall weight that the parent attaches to the welfare of the child is given by the parameter Z > 0. The parent derives utility from the child in two different ways. First, she cares about the child’s actual lifetime utility v, i.e., the parent displays altruism, as in Gary Becker’s work on dynastic utility (e.g., Becker & Barro 1988, Barro & Becker 1989). Second, we also allow for the possibility that the parent may evaluate the welfare of the child through a different function v˜ that does not coincide with the true utility of the child. This feature captures paternalism, i.e., a utility derived from the child’s choices and outcomes that reflects the parent’s own distinct views on what is good for the child. Paternalistic preferences (and the related concept of imperfect empathy) play a central role in the economic literature on cultural transmission.3 The parameter γ measures the relative importance of altruism versus paternalism in the parent’s preferences. We abstract from uncertainty during the two periods of adulthood, but the setup allows for shocks that affect the child before she reaches adulthood. The utility derived from the child is therefore an expected utility. The value function that the child seeks to maximize (and that enters into the parent’s altruistic utility) is v = | + | + . u1 (c1, l1 a1 ) u2 (c2, l2 a2 ) zV 2.

In this case, ct and lt are the consumption and leisure of the child; at is the child’s preference vector at age t;andV is the expected continuation utility of the child after reaching adulthood, which, in a dynastic model, would be of the form in Equation 1. Unlike the parent’s preferences A,the

child’s preferences at evolve during childhood, which captures the gradual formation of attitudes

and noncognitive skills such as patience and perseverance. Leisure lt can be a vector of different activities from which the child derives enjoyment. The parameter z ∈ (0, 1) denotes the weight that the child attaches to future adult utility. Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org

Access provided by Northwestern University on 10/02/19. For personal use only. The parent’s paternalistic concern about the child v˜ is given by v = | + | + ; ˜ u˜1 (c1, l1 A) u˜2 (c2, l2 A) zV

that is, the parent’s paternalistic utility is defined over the same objects as the child’s actual utility but with a potentially different functional form for utility that may depend on the parent’s preferences A. The key implication of the paternalistic component in preferences is that the parent may disagree with the actions of the child; this scope for disagreement turns out to play a central role in the economics of parenting. In particular, parenting styles relate to how the disagreement between parent and child is resolved. Our representation is rather general and can capture different

3Bisin & Verdier (2001) provide a seminal contribution to this literature; Bisin & Verdier (2011) provide a survey.

58 Doepke • Sorrenti • Zilibotti EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

sources of disagreement, ranging from different views regarding risk aversion or work ethic to the consumption of particular goods such as alcohol and cigarettes. A particularly common source of disagreement between parents and children (which we focus on below) stems from time preferences: The parent may wish the child to focus more on long-run success rather than instant gratification. More formally, the parent (in the paternalistic utility v˜ ) places relatively more weight on the child’s future utility V (compared to the young-age utility) than does the child in her own utility function v. Notice that, even though the same discount factor z appears in v˜ and v, our setup still incorporates the possibility of disagreement about intertem- · · poral trade-offs through the dependence of the utility functions ut ( )andu˜t ( ) on the preferences A and a of the parent and the child. As a specific example, we can capture such a difference in patience by setting the utility of the child to

v = + β + β2 a1u(c1, l1 ) a2 u(c2, l2 ) V ,3.

whereas the paternalistic preference of the parent is given by

v = + β + β2 ˜ u(c1, l1 ) u(c2, l2 ) V ,4. √ wherewedefineβ ≡ z. In this case, β is the parent’s discount factor, whereas the child’s (scalar) ≥ ≥ preference parameters a1 1anda2 1 capture additional weight that the child places on experiences early in life relative to the later-in-life expected utility V .4 Next, we turn to choices and constraints. The parent’s choices at time t include her own con-

sumption Ct , the child’s consumption ct , her own leisure Lt , and her own labor supply Nt .Weare particularly interested in parenting choices, i.e., choices aimed at affecting the development of the

child. The parent’s investment in the child’s development is represented by a vector It ,where = { }. It Xt , Et

In this case, Xt is a time investment in child rearing, and Et is a monetary investment (expense) in child rearing. The parent can also affect the influence of peers on the child’s development by

choosing the district of residence dt where the family lives. Finally, the parent can determine the X choice set t for the child, which determines how much freedom of choice the child has. The parent is constrained by an intertemporal budget constraint, + + + + 1 + + + = w + 1 w C1 c1 E1 q(d1 ) C2 c2 E2 q(d2 ) 1 (S) N1 2 (S) N2,5. R R

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org and by a time constraint, Access provided by Northwestern University on 10/02/19. For personal use only. + + = . Lt Nt Xt 1 w Here, R is the gross interest rate, S is the skill vector of the parent, t (S) is the wage, and q (dt ) is the rental cost of living in neighborhood dt . Neighborhoods may differ for exogenous reasons, such as the quality of public schools. There may also be peer externalities, namely, the quality of neighborhood is a function of the average skills of the residents. If households cannot borrow, then the intertemporal budget constraint is replaced by two sequential constraints.

4 | = | = These utilities can be mapped into Equations 1 and 2 by setting u1 (c1, l1 a1 ) a1u (c1, l1 ), u2 (c2, l2 a2 ) β | = | = β a2 u (c2, l2 ), u˜1 (c1, l1 A) u (c1, l1 ),andu˜2 (c1, l1 A) u (c2, l2 ). In Section 3.2, we focus on the case where = > = 2 a2 a 1anda1 a . This case yields a constant intertemporal trade-off, where the child’s discount factor β/ β v = 2 + + 2 a is smaller than the adult discount factor . In this case, we can write a u(c1, l1 ) bu(c2, l2 ) b V v = + β + β2 = β/ <β and ˜ u(c1, l1 ) u(c2, l2 ) V ,whereb a .

www.annualreviews.org • The Economics of Parenting 59 EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

The child’s choices are leisure lt and a vector of own investment in skills xt . The choices of the child are constrained by the choice set imposed by the parent: { }∈X . lt , xt t

The choice set reflects time constraints as well as additional restrictions potentially imposed by X the parent. In turn, the parent’s choice of t has to respect technological constraints, i.e., there is X X ∈ X X a set of feasible choice sets t ,andtheparentmustsatisfy t t . The feasible set t may in turn depend on the investments It , capturing the idea that restricting or expanding the child’s choice may be costly (e.g., time spent monitoring the child). For example, the parent may spend effort to force the child to study more for school and to spend less time playing video games, or the parent may force the child to adopt a particular occupation (e.g., accountant instead of circus performer). Conversely, the parent may expose the child to role models or activities that broaden the range of actions among which the child can choose.5 The final set of constraints for both parent and child concerns the technology of skillaccu-

mulation. The child’s skills consist of cognitive skills ht and noncognitive skills (e.g., preferences, ={ } attitudes, and values) at .Letst ht , at denote the entire skill vector of the child. Let s1 denote the initial skill vector of the child. In the first period of childhood, the child’s skills evolve asa

function of initial conditions, the parent’s investment, the environment, and the child’s effort x1: = . s2 f1(S, s1, I1, d1, x1 ) 6.

In the second period (adolescence), we have = S f2(S, s2, I2, d2, x2 ), 7.

where S ={H , A} is the child’s skill vector at the beginning of adulthood, which in turn determines the child’s utility as an adult V . We now have all the pieces in place to discuss the three specific elements of the economics of parenting outlined in the introduction (Section 1). In Section 3, we focus on parenting style, which relates to how the disagreement between parent and child is resolved. In Section 4, we deal with the technology of skill formation, which relates to the specific properties of the technology in Equations 6 and 7 and how these properties interact with other aspects of parenting. The role

of the choice of a neighborhood dt in parenting is analyzed in Section 5.

3. THE ECONOMICS OF PARENTING STYLE Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org

Access provided by Northwestern University on 10/02/19. For personal use only. The framework outlined in Section 2 can shed light on a variety of aspects of parent–child interactions. In developmental psychology, starting with the seminal work of Baumrind (1967), the concept of parenting style takes a central place in categorizing types of parenting. For us, the choice of parenting style describes the extent to which parents interfere with their children’s own inclinations and how they go about getting children to conform to their own wishes.

3.1. Parenting Styles in the General Model = In the general framework above, consider the child’s choices xt in period t 1, 2, which may include actions such as effort put into studying for school, but also other choices with potential

5The parent’s control over the child’s choice set relates our model to Bandura’s (1986) social cognitive theory. According to this theory, children learn from role models to which they are exposed. Saez Marti (2018) incorporates this notion into an economic model of cultural transmission through endogenous parental influence.

60 Doepke • Sorrenti • Zilibotti EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

long-run consequences, such as drinking alcohol, having unprotected sex, and similar risky behaviors. In the analysis of Doepke & Zilibotti (2017), parenting styles come down to whether, and

how, parents interfere with the child’s choice. If the child can choose xt freely, then she will do this with the objective of maximizing her own utility in Equation 2. How will the parent feel about this choice? If the parent is purely altruistic (i.e., if γ = 0, so that only the altruistic component enters Equation 1), then the parent will necessarily agree. Such a parent simply wants the child to be happy and thus approves of anything that contributes to the child’s happiness. However, if the parent is at least somewhat paternalistic (γ>0), then there may be disagreement. For example, if the parent places relatively more weight on the child’s future relative to the enjoyment of the present, then the parent will wish the child to put more effort into education than what the child would choose of her own accord. Similarly, if the parent’s paternalistic preferences incorporate more risk aversion than the child’s own utility, the parent may wish the child to stay away from certain risk-taking activities that the child would like to engage in. The three traditional parenting styles in developmental psychology can be understood as archetypes of resolutions of this conflict between parent and child.6 The permissive parenting style is one where the parent lets the child have her way and refrains from interfering in the

choices xt . The authoritarian style is one where the parent imposes her will through coercion. In the model above, coercion (i.e., authoritarian parenting) is captured through the notion of re- X X stricting the choice set t . An authoritarian parent chooses a small set t that leaves little or no leeway to the child. The third parenting style, authoritative parenting, is also one where the parent aims to affect the child’s choice. However, rather than using coercion, an authoritative parent

uses persuasion: She shapes the child’s preferences a2 through investments I1 in the first period of life. For example, such a parent may preach the virtues of patience or the dangers of risk when the child is little, so that the child ends up with more adult-like preferences when the child’s own decisions matter during adolescence. What determines the trade-off among these parenting styles? All parenting styles come with costs and benefits. Permissive parenting maximizes the utility of the child, but the child mayend up making choices that go against the paternalistic drive of the parent. An authoritarian parent directly influences the child’s choice, but this lowers the happiness of the child and may haveother repercussions, particularly if there are economic benefits (in terms of the child’s future expected utility) to making independent decisions. An authoritative parent also influences the child’s choice, but this parenting style requires costly effort on the part of the parent and may also reduce the child’s welfare. How these trade-offs are resolved depends in part on the level of paternalism γ . More importantly, for a given level of paternalism, what matters are the economic conditions that

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org shape the impact that a given decision will have on the child’s future. Access provided by Northwestern University on 10/02/19. For personal use only.

3.2. The Impact of Economic Conditions on Parenting Style We illustrate the interaction between economic conditions and parenting styles with a specialized version of the model in Section 2 in which the dimension of child preferences that parents aim to influence is patience.7 Patience is a natural starting point for a theory of parent–child transmission of preferences, both because disagreement about how to weigh immediate gratification versus future payoffs is a particularly salient aspect of parent–child interactions and because patience

6Neglecting parenting is sometimes considered to be a fourth parenting style. We abstract from this parenting style in this review (for a discussion, see Doepke & Zilibotti 2017). 7Models of endogenous patience were first applied by Becker & Mulligan (1997) and Doepke & Zilibotti (2005, 2008).

www.annualreviews.org • The Economics of Parenting 61 EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

matters for future-oriented investments such as education.8 The stylized model is based on the work of Doepke & Zilibotti (2017) and Agostinelli et al. (2019). The model is stripped down to the essential elements needed to bring out the main trade-offs involved in choosing a parenting style. We abstract from goods consumption and labor supply. For the parent, the period utility function = − is linear in leisure, given by L1 1 X1,whereX1 is parenting effort. There is no parenting effort in the second period, so that (ignoring constants) the utility in Equation 1 takes the form − + − γ v + γ v . X1 Z [(1 ) ˜ ] For the child, we abstract from utility during the first (early childhood) period. During the second

(adolescence) period, the child gets linear utility from leisure l2, which depends on the child’svector x2 of education investments. There is disagreement about the intertemporal trade-off between the child and the paternalistic parent, as in Equations 3 and 4. Given that utility is only derived in the second period, we can rewrite Equations 3 and 4 as v = + β a2l2 V ,8. v = + β ˜ l2 V ,9.

≥ where a2 1 is the extent to which the child places more weight on fun in the present versus welfare in the future. Next, we turn to the technology of skill formation. In the first period, the technology in ={ } Equation 6 for the accumulation of the skill vector s2 h2, a2 takes the form = h2 fh,1 (S, d1 ), = . a2 fa,1 (S, X1, d1 )

That is, the child’s cognitive skills in adolescence h2 evolve passively depending on endowments (represented by the parent’s skill vector S) and the environment (represented by the neighborhood

quality d1, taken as fixed in this case). We abstract from investment in cognitive skills byboththe parent and the child at this stage. Instead, parental involvement is crucial for the acquisition of

noncognitive skills, represented in this case by the child’s preference parameter a2. In addition to S and d1, this parameter depends on the parent’s time investment X1. Intuitively, increasing the child’s patience (in this case, by lowering the enjoyment of the present a2) requires a socialization X effort that is costly in terms of time. In the first period, the parent also chooses the choice set 2 that constrains the child’s choice in the second period.

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org In the second period (adolescence), adult skills are formed. At this stage, parents make no fur- Access provided by Northwestern University on 10/02/19. For personal use only. ther direct investment, and the ball is in the child’s court. The child is now called on to make

decisions x2 in two dimensions: the choice of an occupation k and an investment in human capital ={ } = xh,2, i.e., x2 k, xh,2 . The acquisition of adult skills S H hinges on xh,2 and the values and skills that the child has acquired in the early period: = . H fh,2 (S, h2, d2, xh,2 )

In this case, H is strictly increasing in the educational effort xh,2.

8Figlio et al. (2016), Heckman et al. (2006), and Sutter et al. (2013) provide evidence at the micro level, and Dohmen et al. (2018) provide a country-level analysis of the relationship between patience and development. Doepke & Zilibotti (2017) also consider the transmission of risk preferences, which have been shown by Dohmen et al. (2012) to be strongly correlated between parents and children. Zumbuehl et al. (2013) find that risk attitudes are particularly correlated between parent and child when parents invest a lot in child-rearing effort, which is consistent with the general model of parenting described in this review.

62 Doepke • Sorrenti • Zilibotti EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

The occupations k differ in the relative importance of talent and effort. Specifically, each oc- { w ψ } cupation k is defined by a triplet x¯k, ¯ k, k determining the time x¯k required to learn the basics w ψ of the occupation, the basic wage ¯ k in the occupation, and the return k to human capital H . The potential income that a child would realize in occupation k is given by ξ w + ψ w + − ψ w k ¯ k kH H (1 kH ) L ,

ξ where k is an independent and identically distributed individual talent shock capturing a Roy- model-type comparative advantage at different occupations with unit mean and finite variance, w > w > w and L 0and H L capture the overall return to education. The talent shock is realized in the second (adolescence) period, before the child decides on x2 (so that the child does not face X uncertainty) but after the parent’s choice of X1 and 2 in the first period. The setup captures the idea that, in some occupations, success hinges primarily on the child’s talent and intrinsic motivation (e.g., becoming an artist or an elite sports player) and thus on her choosing something that she is good at. Such occupations are characterized by a high variance of ξ ψ the talent shock k across individuals but a low return k to formal education. In other occupations, such as becoming an engineer or a business manager, cognitive human capital, and thus success in ξ education, is more important. In such occupations, the talent shock k has little variance or may w ψ even be constant, and the basic wage ¯ k is low, but the return to educational effort k is high. Professions also differ in how difficult it is to learn the basics; children may find it more pleasant

to train to become an actor (low x¯k) than to study math or economics (high x¯k). Overall inequality w > w > w w w is captured by L 0and H L: When the gap between H and L is small, both inequality and the return to education are low.

The child chooses the vector x2 (i.e., the occupation k and educational effort xh,2)soasto maximize her own utility in Equation 8: max a 1 − (x + x¯ ) + zV , ∈X 2 h,2 k x2 2

where = ξ w + ψ w + − ψ w ; V k ¯ k kH H (1 kH ) L

recall that

= . H fh,2 (S, h2, d2, xh,2 ) Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org Access provided by Northwestern University on 10/02/19. For personal use only. Consider how economic conditions affect the choice of parenting style. The basic parent–child conflict underlying this choice is that the parent attaches a higher weight to the future utility V of

the child. As a result, the parent would like the child to exert a greater educational effort xh,2 than the child would do on her own accord. While Doepke & Zilibotti (2017) allow for intensive mar- gins in parenting (e.g., parents can be more or less authoritative and more or less authoritarian), in this section, we simplify the analysis by restricting parents to a discrete choice between the three parenting styles: permissive (PE), authoritative (AV), and authoritarian (AR). Each parenting style ∈{ } P PE, AV, AR comes with a specific effort cost XP for the parent. > The permissive parent PE does not interfere with the child’s preferences, so that a2 1and the child gets full enjoyment of adolescence, and also does not restrict the choice of occupation. In principle, a permissive parent might even spend effort to expand the children’s choice set, for example, by offering more opportunities for entertainment. However, we abstract from this pos- = sibility and set the effort cost of permissive parenting to zero, XPE 0.

www.annualreviews.org • The Economics of Parenting 63 EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

The authoritative parent AV does not restrict the child’s choice set but molds her preferences. = Specifically, this parent socializes the child into adult-like preferences, a2 1. This indoctrination reduces the child’s enjoyment of adolescence while inducing her to make more forward-looking

choices x2, to the extent that the child ends up choosing just what her parent would like her to. > Indoctrinating the child requires effort XAV (S, d1 ) 0, which is decreasing in both the skill of the parent and the quality of the neighborhood. This assumption reflects the idea that better-educated parents possess superior soft skills to persuade their children (e.g., they use more convincing argu- ments to convince them that working for school is a good idea). In addition, the parental effort can be reinforced by a positive neighborhood culture, including good teachers and peers who share work-oriented values. Finally, the authoritarian parent AR forces the child into a particular occupation (i.e., the choice X set 2 given to the child only includes the occupation k preferred by the parent). Authoritarian

parents also impose control on the child’s effort choice xh,2. Imposing these restrictions requires > costly monitoring on the part of the parent, which is captured by a positive effort cost XAR 0. Depending on the parent’s skill and the neighborhood, XAR may be larger or smaller than the cost

of the authoritative style XAV (S, d1 ). Unlike authoritative parents, authoritarian parents do not change their children’s preferences. For them, the problem is not what children think but what they do. Therefore, they force them to do as parents like rather than attempting to persuade them of the validity of the adult standpoint. A downside of the authoritarian style is that the parent locks the child into a particular occupation before the child’s occupation-specific talent shock is realized. As an example of this kind of authoritarian parenting, Doepke & Zilibotti (2017) discuss a setting in which parents force their children to stay at home and work in the family business instead of allowing them to move to the city and study. Doepke & Zilibotti (2019) discuss other forms of control over the process of skill formation. For instance, a young child wishing to become a professional soccer player may be denied the opportunity to practice that sport at an early age. Conversely, a child may be forced to play piano or tennis for many hours so as to induce special- ization to an extent that it delimits the choice set of the child when she grows up. Table 1 summarizes the features of the three parenting styles. We can assess how economic conditions affect the choice between them. Consider, first, a low-inequality society, where the gap w w between H and L is small. In such a society, there is limited incentive for children to put effort xh,2 into education. Parents are also less concerned about children’s effort, and thus there is little scope for disagreement between parents and children. Therefore, most parents adopt a permissive parenting style, namely, they keep young children happy and foster their sense of independence so

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org that they can discover what they are good at in their adult life. There may still be a small fraction Access provided by Northwestern University on 10/02/19. For personal use only. of highly paternalistic parents (high γ ) who want to influence their children’s occupational choice. Among them, those with low skills S will be authoritarian, i.e., resort to restrictions on the choice set. The downside of this parenting strategy is that it prevents children from learning professions in which they have a strong comparative advantage ex post. For instance, talented painters may become not-so-well-paid engineers. Highly paternalistic parents with high skills S will instead be authoritative.

Table 1 Three parenting styles Parenting style Indoctrinate child? Restrict child’s choice? Cost to parent > Permissive No (a2 1) No 0 > > Authoritarian No (a2 1) Yes XAR 0 = > Authoritative Yes (a2 1) No XAV (S, d1 ) 0

64 Doepke • Sorrenti • Zilibotti EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

Consider, next, a high-inequality society. There, the disagreement between parents and children is more salient because parents would like to see their children work hard in school and choose professions with a high return to human capital. In this society, a larger share of parents will be authoritative, and fewer will be permissive. Moreover, because of the comparative advantage of rich and educated parents in authoritative parenting, there will be a stronger socioeconomic sorting into parenting styles. Since an authoritative parenting style is conducive to more economic success, this sorting will hamper social mobility. While inequality drives the choice margin between permissive and authoritative parenting style, other factors determine the appeal of authoritarian parenting. Doepke & Zilibotti (2017) discuss the role of the return to incumbency. In traditional societies, most children adopt their parents’ profession and learn the necessary skills inside the family. Thus, parents have little incentive to be permissive to let children discover what they are good at. Nor do they need to spend effort in socializing children into adult-like values (i.e., to be authoritative), since they can achieve the same result by simply monitoring them. As the economy develops, occupational mobility grows, the return to incumbency falls, and the appeal of authoritative parenting declines. In our model, we can parameterize the value of independence by the variance of the talent ξ shock k. When this variance is low, the opportunity cost of authoritarian parenting is small. As ξ the variance of k increases, so does the value of letting children discover their talent. Thus, fewer and fewer parents will be authoritarian and restrict their children’s occupational choice.

3.3. Empirical Evidence on Economic Conditions and Parenting Styles The discussion in Section 3.2 generates several testable predictions. Doepke & Zilibotti (2019) show that, in many industrialized countries, the trend of rising income inequality over recent decades has been accompanied by more intensive (authoritarian or authoritative) parenting, while permissive parenting has declined. Evidence from the Multinational Time Use Study shows that, in all countries for which a sufficient number of observations is available (Canada, Italy, the Netherlands, Spain, the United Kingdom, and the United States), time diaries display a significant increase in the time that parents spend with their children over the past few decades. For the United States and the Netherlands, we have long time series, which makes these two cases especially informative. In both countries, college-educated parents have increased the time spent on child care much more than have high-school graduates. In the 1970s, low- and high-educated parents in the United States spent approximately the same amount of time on child care. Today, there is a gap of more than three hours per week between more and less educated parents. Ramey

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org & Ramey (2010), who were the first to note this diverging behavior, suggest that this trendis Access provided by Northwestern University on 10/02/19. For personal use only. explained by increased competition for admission to top colleges. A large share of the increase in child care time is related to educational activities. In 1976, US couples spent an average of two hours a week on playing with, reading to, and talking to their children and approximately 17 minutes a week on helping them with homework. In 2012, the average went up to six and a half hours a week for playing, reading, and talking to children and more than 1.5 hours for helping them with homework. Overall, US parents now spend 3.5 times more time on these education-related child care activities. This shift to more intensive and less permissive parenting styles is reflected in children’s experiences. The percentage of kids walking or biking alone to school fell from 41% in 1969 to 13% in 2001. Among 6-to-8-year-old US children, unsuper- vised play time decreased by 25% between 1981 and 1997, whereas time spent on homework more than doubled. This is consistent with parents pushing children toward academic achievement. This rise in child care time is associated with the widely discussed phenomenon of helicopter parenting, i.e., the trend (especially pronounced among the upper and middle classes) toward an

www.annualreviews.org • The Economics of Parenting 65 EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

increasingly intrusive and protecting parenting style, with a mix of authoritarian and authoritative elements. Below, we relate these trends to changing macroeconomic conditions in areas such as income inequality, the return to education, and redistributive policies, following the investigation of Doepke & Zilibotti (2017, 2019). First, we document that parents are less permissive (and, especially,more authoritative) in countries with higher inequality across Organisation for Economic Co-operation and Development (OECD) countries.9 We use information from the World Value Survey (WVS), in which people are asked to select five values (out of a list of ten) that children should be encouraged tolearn at home.10 To simplify the discussion of the evidence, we use a binary classification of parenting styles. We classify as intensive parents who single out either obedience (a value typically associated with an authoritarian parenting style) or hard work (which is associated with an authoritative parenting style). We classify as relaxed those who mention either independence or imagination and neither obedience nor hard work. Intensive parenting corresponds to a combination of authoritarian and authoritative methods, and relaxed parenting corresponds to permissive methods.11 This classification covers 91% of a sample of 66,632 respondents from OECD countries. Among them, 63% are classified as intensive, and 37% are classified as relaxed. We exclude from theanal- ysis the 9% of respondents who remain unclassified because they do not mention any of the four above-mentioned values. We study the correlation between intensive parenting and measures of pretax earnings inequality and government-mandated redistribution. In particular, we use (a) the 90th-to-10th percentile ratio in the pretax earnings of full-time dependent employees, (b) the return to tertiary education (Montenegro & Patrinos 2014), (c) a measure of tax progressivity that captures the marginal rate progression up to an income level equivalent to four times a country’s per-capita GDP (Andrew Young School Policy Stud. 2010), and (d) a measure of aggregate social expenditure as a percentage of GDP from the OECD Social Expenditure Database.12 Figure 1 plots the results for Wave 5 of the WVS, which was carried out in 2005 and has the largest number of countries. As predicted by the theory, the share of intensive parents increases with pretax inequality and the return to education and decreases with the extent of redistribution through tax progressivity and social expenditure. For instance, 79% of US respondents are classified as intensive—and the United States has both high earnings inequality and a high return to education and low tax progressivity and social expenditure. Among the US respondents, 49% are authoritative, and 30% are authoritarian in the classification of Doepke & Zilibotti (2017). In contrast, only 26% of the Swedes are intensive parents, whereas 74% are relaxed—and Sweden has low inequality and high redistribution. Countries with moderate inequality, such as Germany

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org and Japan, fall in between. The overall pattern conforms well with the prediction of the theory Access provided by Northwestern University on 10/02/19. For personal use only. for all measures of inequality.

9The sample includes all countries with available data that were OECD members as of 1994: Australia, Canada, Finland, France, Germany, Italy, Japan, the Netherlands, New Zealand, Norway, Spain, Sweden, Switzerland, Turkey, the United Kingdom, and the United States. The cross-country pattern is robust if we include the nine countries with available data that became OECD members after 1994 (Chile, the Czech Republic, Estonia, Hungary, Israel, Korea, Mexico, Poland, and Slovenia). 10The question is worded as follows: “Here is a list of qualities that children can be encouraged to learn at home. Which, if any, do you consider to be especially important? Please choose up to five!” The values among which parents can choose are independence, hard work, feeling of responsibility, imagination, tolerance and respect for others, thrift and saving money, determination and perseverance, religious faith, unselfishness, and obedience. 11Doepke & Zilibotti (2017) show that the results are robust to other classifications, such as defining as authoritative all nonauthoritarian parents who mention either hard work or determination and perseverance among thetopfivevalues. 12Note that the third and fourth measures are inverse measures of inequality. 66 Doepke • Sorrenti • Zilibotti EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

1.0 1.0 Turkey Turkey Spain Spain United France France States 0.8 United States 0.8 Great Britain Canada New Australia Australia Canada Italy Zealand 0.6 Netherlands Italy Netherlands 0.6 Norway 0.4 Germany Finland Japan Switzerland Germany 0.4 Sweden Norway Japan 0.2

Share of intensive parents of intensive Share Sweden parents of intensive Share 0.2 0.0 2.0 2.5 3.0 3.5 4.0 4.5 5.0 6 9 12 15 18 Inequality (decile 9/1) Returns tertiary education

1.0 1.0 Turkey Turkey Spain Spain France France 0.8 United States 0.8 United States Canada Great Britain Canada Great Britain Australia Australia Italy Netherlands Netherlands 0.6 New 0.6 New Italy Zealand Zealand Finland Finland Norway Switzerland Norway 0.4 Germany 0.4 Germany Switzerland Japan Japan Sweden Sweden Share of intensive parents of intensive Share parents of intensive Share 0.2 0.2 0.02 0.05 0.08 0.11 0.10 0.15 0.20 0.25 0.30 Tax progressivity Social expenditure (percent of GDP)

Figure 1 Inequality, redistribution, and intensive parenting across countries.

The cross-country results still leave open the possibility that parenting styles are mostly due to some unchanging local tradition, and that the correlation with inequality is due to omitted variables. It is therefore important to verify the results using changes in inequality and parenting styles within countries over time. In this section, we focus on changes in pretax earnings inequality, for which we have information at a higher frequency. We expect that intensive parenting increases in countries where inequality increases and decreases in countries that become less unequal. Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org

Access provided by Northwestern University on 10/02/19. For personal use only. To examine this prediction, we perform regression analysis based on individual data. We estimate the following equation:

= α + α + β · + · γ + ε INT_PARict c t INEQct Xict ict ,

where i, c,andt stand for individual, country, and time (wave). The dependent variable INT_PAR is an indicator for parenting style, where INT_PAR = 1 means that the parent is intensive, whereas = α INT_PAR 0 means that the parent is relaxed. Among the right-hand side variables, c is a coun- α try fixed effect, t is a wave fixed effect, INEQct is a time-varying measure of inequality (the 90–10 earnings ratio), X is a vector of individual and country characteristics including gender, age, age squared, and the (log of ) GDP, and ε is an error term. We estimate a logistic regression on 45,482 individual observations. Table 2 displays the estimates for β, expressed as odds ratios. All regressions include wave fixed effects with standard errors clustered at the country level. Column 1 displays the basic specification. Column 2 adds

www.annualreviews.org • The Economics of Parenting 67 EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

Table 2 Inequality, redistribution, and parenting styles (1) (2) (3) (4) (5) (6) Logit Logit Logit Logit Logit Logit intensive intensive intensive intensive intensive intensive ∗∗∗ ∗∗∗ ∗∗ ∗∗∗ ∗∗∗ ∗∗ Inequality 2.38 2.50 2.12 1.74 1.74 27.22 (0.44) (0.29) (0.72) (0.37) (0.28) (35.21) ∗∗ ∗∗ Tax progressivity 0.20 0.24 5.35 (0.13) (0.17) (5.88) ∗∗ Social expenditure 0.70 0.58 0.21 (0.29) (0.25) (0.14) Controls No Yes Yes No Yes Yes Country fixed effects No No Yes No No Yes Observations 45,482 45,482 45,482 32,196 32,196 32,196

The dependent variable is intensive parenting style (indicator). All the models are logistic models, and the displayed coefficients are odds ratios. All the models include wave fixed effects. Models in columns 2, 3, 5, and 6 include controls for gender, age, age squared, and log of GDP per capita (basedon expenditure-side real GDP at chained purchasing power parities, from Penn World Table 9.0). Models in columns 3 and 6 also include country fixed effects. Inequality is defined as the ratio between the 90th and 10th percentile of gross earnings of full-time dependent employees.yisfrom Taxprogressivit the Andrew Young School of Policy Studies (Andrew Young School Policy Stud. 2010). Safety nets are expressed as the aggregate social expenditure as a percentage of GDP. Standard errors (in parentheses) are clustered at the country level. ∗, ∗∗, ∗∗∗ indicate statistical significance at the 10%, 5%, and1% level, respectively.

the control variables Xict . Column 3 also adds country fixed effects. The odds ratio is significantly larger than unity and stable across specifications. In all cases, higher inequality increases the probability of intensive parenting. In columns 4, 5, and 6, we repeat the analysis while also including the measures of tax progressivity and social expenditure. The results in columns 4 and 5 confirm that each of the three variables of interest has the predicted effect: Inequality increases the intensive parenting style, while tax progressivity and social expenditure reduce it (although the effect of the latter is not statistically significant). In the regression in column 6, the three effects areless well identified. The effects of inequality and social expenditure conform with the prediction of the theory, but the odds ratios are very high and low, respectively. Tax progressivity is not statistically significant. The results in the regression in column 6 reflect the limited time variationfor the measures of redistribution, which makes it difficult to obtain reliable estimates in fixed-effect regressions.13 The correlations are quantitatively large. Consider, for instance, a hypothetical parent born in a country where the 90-to-10 measure of inequality is three, close to the average for European countries. Suppose that the individual characteristics of this parent and the country fixed effect are Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org

Access provided by Northwestern University on 10/02/19. For personal use only. such that her probability of being intensive is 40% (approximately the average value for countries like Germany, Japan, and Switzerland). Now consider an increase of inequality from three to five, close to the US value, holding other characteristics—including the country fixed effect—constant. Then the probability for this parent to be intensive increases from 40% to 75%. Although this estimated change cannot be interpreted as causal, this is a large effect, moving the likelihood of being an intensive parent close to the proportion of intensive parents in the United States. Above, we focus on the relaxed versus intensive margin (or, equivalently, permissive versus nonpermissive), showing that high inequality induces parents to become less relaxed. Consider next the margin of authoritarian versus nonauthoritarian parenting (i.e., parents who emphasize

13All results hold true irrespective of whether we include all respondents or only those who are actually parents. They are also robust to the inclusion of a self-reported dummy for religiosity. Finally, they are robust to an alternative definition of parenting style, where we classify as intensive parents who mention either hardwork or determination and perseverance among the preferred parenting values.

68 Doepke • Sorrenti • Zilibotti EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

GHA 0.8 RWA 0.8 RWA TTO IRQ EGY BFA BFA EGY ZMB COL JOR COL JOR MEX THA MEX THA PER PER 0.6 BRA IND 0.6 IND CHL BRA CHL POL IDN MAR POL IDN UKR GBR CYP UKR MAR GBR ZAF TUR CYP SRB ARG ARG SRB NLD URY TUR URY NLD FRA ESP ESP 0.4 ETH 0.4 ETH FRA AUS AUS FIN RUS RUS FIN CAN HUN SVN IRN IRN HUN SVN USA USA NOR NZl VNM ITA ITA MYS MYS BGR NOR NZL BGR GEO GEO CHE CHE 0.2 DEU ROU 0.2 ROU MDA CHN SWE Share of authoritarian parents Share SWE CHN of authoritarian parents Share MDA KOR KOR JPN HKG JPN 0.0 HKG 0.0

0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 1.0 Employment share of agriculture Enrollment rate in tertiary education

Figure 2 Share of agriculture, higher education, and authoritarian parenting.

obedience versus those who do not). In this case, the key variables are (a) whether the effort of children can be easily monitored and (b) the extent of occupational mobility (which affects the return to independence and thus the choice between authoritarian and authoritative parenting). Two (highly correlated) proxies for macroeconomic conditions that affect this margin are the share of agriculture and the enrollment rate in tertiary education. In traditional agricultural societies, most children live and work under the parents’ direct supervision. Moreover, most of them are expected also to become farmers and to learn skills from their parents. In such societies the return to independence is low. Conversely, in societies where many children receive higher education, they are expected to work hard on their own initiative, as students often live far from their families and must make important independent decisions, such as which subject to major in and how hard to study. Therefore, we should expect parents to be more authoritarian in traditional agricultural societies where the share of enrollment in higher education is low. Figure 2 shows that these predictions are borne out in the data: Across countries, the share of authoritarian parents is increasing in the employment share of agriculture and decreasing in the enrollment rate in tertiary education. Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org

Access provided by Northwestern University on 10/02/19. For personal use only. Doepke & Zilibotti (2019) also examine the effect of religiosity on parenting. They show that religious people are significantly less permissive and more authoritarian than nonreligious people. This finding is consistent with the hypothesis that many traditionally religious parents believe that the world is regulated by a never-changing order, and that it is their duty to trans- mit to their children an immutable set of values and truths.14 The result is based on US data from the Child Development Supplement (CDS) of the Panel Study of Income Dynamics (PSID). In the initial survey of the PSID-CDS in 1997, parents are asked whether they have ever spanked their children. In addition, they are asked whether religion is very important in their life. The data set also contains information on parenting styles. Religious parents have a more positive

14According to Bartkowski & Ellison (1995, p. 25), such parents do not emphasize creativity and intellectual curiosity, but instead believe that, “to succeed in adult roles, children must be trained to embrace the divinely- ordained principles of authority and hierarchy.” Therefore, religious parents may be more prone to being authoritarian.

www.annualreviews.org • The Economics of Parenting 69 EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

attitude toward corporal punishment: 70% of religious parents report spanking their children, compared to 58% of nonreligious parents. These results are robust to controlling for a variety of socioeconomic characteristics, such as income or education, that may be related to both religiosity and the use of corporal punishment. In addition, nonreligious parents are significantly more permissive than religious ones, especially Protestants. The effect is quantitatively large. Consider a nonreligious parent whose probabilities of being permissive, authoritative, and authoritarian are, respectively, 50%, 25%, and 25%. The probabilities that an otherwise identical parent who is an observant Protestant will embrace each of the three parenting styles are, respectively, 30%, 25%, and 45%. Similar patterns are observed in a cross-country analysis based on the WVS. Overall, the evidence lines up very well with the predictions of our simple model of the economics of parenting. Empirically, economic factors can account for a substantial fraction of the variation in parenting practices over time and across space. Clearly, the results do not rule out that noneconomic factors also matter, as in the role of religiosity discussed above. Also, the evidence is at the level of correlations, and more work toward causal identification is needed to confirm the results. Nevertheless, the results are indicative that the economic approach can be quite successful in explaining what parents do.

3.4. Additional Mechanisms for the Choice of a Parenting Style The economic literature on parenting has identified additional mechanisms that are relevant for the choice of a parenting style but that are not captured by the model outlined in Section 3.2. Among the earliest contributions is that of Weinberg (2001), who focuses on the role of monetary incentives in raising children. In his model, richer parents can use monetary rewards to get their children to comply with their wishes. Poor parents lack the resources to do the same and thus may be more likely to resort to authoritarian methods such as corporal punishment. Thus, the mechanism can contribute to our understanding of the distribution of parenting styles in the population and also help explain the impact of parental income on child achievement. Another interesting dimension is information frictions. At least initially, children know a lot less than parents do, and part of the challenge of parenting is to convince children of what the parents know is right. Information transmission from parent to child may be distorted for several reasons; for example, the child may mistrust the parent and believe that she is being fed wrong information for strategic reasons (“If you don’t eat your vegetables, you will never grow big!”), or it may simply be difficult to convey complex, tacit, or informal knowledge. In their model,

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org Lizzeri & Siniscalchi (2008) focus on the polar case where the parent is unable to directly convey Access provided by Northwestern University on 10/02/19. For personal use only. information, and the child learns entirely from experience. The parent faces a trade-off between letting the child learn from her own mistakes or correcting the child at the cost of inhibiting learning. As in the model above, how much is at stake in these decisions matters for the optimal parenting strategy.In a recent paper, Seror (2019) combines the principal-agent approach followed by Lizzeri & Siniscalchi (2008) with some of the trade-offs incorporated in the model described in Section 3.2. Other recent studies involving a choice of parenting strategies include those of Burton et al. (2002), who focus on how a parent responds to bad behavior by a child; Cobb-Clark et al. (2019), who introduce a distinction between parents’ time and attention as inputs in child rearing; Cosconati (2013), who considers the contrast between a strict and a permissive parenting style in a setting where parent and child disagree on the relative importance of the child’s leisure; Zhang & Ikeda (2016), who let a central role be played by a process of habit formation of which only the parent is fully aware; Hao et al. (2008), who focus on strategic interplay between strict and lenient

70 Doepke • Sorrenti • Zilibotti EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

parenting in families with multiple children; and Lundberg et al. (2009) and Romich et al. (2009), who analyze the noncooperative interaction between paternalistic parents and myopic children. Cunha (2015) considers the role of information frictions on the part of parents, who may differ in their subjective assessment of the productivity of different parenting styles. The self-interest of parents takes center stage in the work of Becker et al. (2016), where parents may attempt to manipulate their children’s preferences to receive more support in old age. None of these studies attempt to account for variation in parenting styles across countries and over time. Instead, they introduce a broad range of mechanisms and forces that are potentially relevant in the choice of a parenting style and that could be included in more encompassing future models of parenting behavior.

4. THE ECONOMICS OF CHILDREN’S SKILL ACQUISITION Skills and abilities are important determinants of an individual’s life and economic success. In economics, the process of skill acquisition first took center stage in the early literature onhuman capital that was pioneered by Gary Becker and Jacob Mincer, among others. Initially, the analysis of human capital accumulation was focused on cognitive skills that traditionally have been deemed relevant for shaping outcomes such as education and earnings (Murnane et al. 1995, Cawley et al. 2001). More recent literature has also shown how noncognitive skills such as perseverance, patience, sociability, and other personality and behavioral traits are important determinants of outcomes in childhood and adulthood (Heckman & Rubinstein 2001; Heckman et al. 2006, 2013). Different disciplines such as economics, psychology, and sociology share the view that childhood is a critical period in an individual’s life because this is when many of the attributes that will affect adult outcomes are acquired. In economics, the consideration of family influences during childhood was popularized by the work of Becker & Tomes (1979, 1986), who analyze the role of endowments versus investment in the determination of intergenerational mobility. Since then, a large literature has aimed to understand the process through which children acquire skills in detail. In our parenting model in Section 2, the technology of skill acquisition is represented by Equations 6 and 7, reproduced here for easy reference: = s2 f1(S, s1, I1, d1, x1 ), = S f2(S, s2, I2, d2, x2 ),

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org

Access provided by Northwestern University on 10/02/19. For personal use only. where the skill vectors during childhood st and adulthood (S for the parent, S for the child) include both cognitive and noncognitive skills. This formulation of skill acquisition already incorporates some central insights of the recent literature on skill acquisition. For example, the technology gives a role to both endowments (represented by the parent’s skills S and the child’s initial skills

s1) and later investments in skills. The technology also allows for investment on the part of both

parent (It ) and child (xt ), and the environment (such as schools, neighborhoods, and peers) also matters (dt ). An important feature of this formulation is that it allows for the cumulative formation of skills. The skills h2 acquired during childhood are one of the inputs in skill acquisition during adolescence. The model allows for different degrees of complementarity or substitutability of early and late investments in skills. A focal point of the literature on skill formation is to pin down the technology in Equations 6 and 7 in more detail, beyond the general functional form. Some of the central questions in this literature are:

www.annualreviews.org • The Economics of Parenting 71 EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

What is the relative importance of endowments, parental investment, own investment, and the environment for the acquisition of cognitive and noncognitive skills? During which phase of childhood are investments in cognitive and noncognitive skills most effective? Are early and late investments in skills complements or substitutes? All of these questions are relevant for parents who want to decide which parenting style to adopt and how to best invest in the skills of their children. What is more, these questions are also crucial for the design of public policy. The timing of skill acquisition helps determine at what ages public investments in school and child care are likely to be most effective. The issue of substitutability versus complementarity matters for whether early deficits (say, regarding children from socially disadvantaged backgrounds) can be offset by interventions later on (say, training and tutoring during high school) or should be addressed as early as possible. The substitution relationship between parental and other inputs (such as public schooling) determines whether public programs that aim to foster skill acquisition can concentrate on providing public inputs, or whether support and tutoring for parents should be a central plank of such programs. A large body of research during the past two decades provides answers to many of these questions. Recent comprehensive overviews of this literature are provided by Heckman & Mosso (2014) and Attanasio (2015). We focus on the main insights of this literature and some of the most recent developments. The work by James Heckman and his coauthors represents a central part of our understanding of the process of skill formation. Building on the framework of Becker & Tomes (1979, 1986), this research describes the complex process of skill formation through the formalization of a multistage technology of skill formation. The combination of a theory for the technology of skill formation (Cunha & Heckman 2007) with estimates of the skill production function (Cunha & Heckman 2008, Cunha et al. 2010) makes it possible to characterize the process of human development and to rationalize the available empirical evidence. First, as reflected in the technology in Equations 6 and 7, skills are multidimensional (Almlund et al. 2011). Cognition and IQ are not the only determinants of success in life; personality traits such as perseverance, conscientiousness, and sociability are equally important in shaping future outcomes. Second, skills are malleable, and genetics is only one of the main possible determinants of the set of individual skills and abilities. Perhaps the most important finding in the recent literature is that early childhood (up toap- proximately four years of age) is a crucial phase for skill acquisition. Carneiro & Heckman (2003), Ermisch & Francesconi (2005), and Heckman & Mosso (2014) document that cognitive and noncognitive outcomes are largely shaped early in life.15 The skill gaps that can be found among Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org

Access provided by Northwestern University on 10/02/19. For personal use only. individuals from different socioeconomic backgrounds opens at very early stages of childhood. Moreover, early interventions have higher returns and are more effective in reducing achievement gaps than later interventions. Later investments are especially ineffective if not anticipated by at least some earlier investments (see Cunha et al. 2006, Knudsen et al. 2006). Our theoretical representation of the technology of skill formation in Equations 6 and 7 is closely related to that of Cunha & Heckman (2007). A key feature is that childhood consists of multiple periods.16 The model allows for two important features. The first feature is self-productivity of human capital. Skills acquired in a certain period persist and contribute to the creation of new

15Works such as those of Carneiro & Heckman (2003), Heckman & Masterov (2007), and Cunha et al. (2006) show the important role of parents—through ability and engagement—in producing both cognitive and noncognitive skills. 16The recognition of different stages of child development was widely acknowledged in the child development literature (Erikson 1950) but initially neglected by the literature on the economics of child development.

72 Doepke • Sorrenti • Zilibotti EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

skills in the following periods. The second feature is dynamic complementarity, namely, the observation that skills produced at a specific stage of life raise the productivity of investments atlater stages. This implies that investments made at different ages foster each other. Moreover, complementarity suggests that early-in-life investments should be followed by later investments to increase their productivity. Empirical evidence on these features is provided by Cunha & Heckman (2008) and Cunha et al. (2010). Cunha & Heckman (2008) estimate a linear dynamic factor model to identify the multistage investment technology under the assumption that early and late investments are per- fect substitutes. Cunha et al. (2010) identify a more general nonlinear technology. Estimates are obtained through a dynamic factor model where both cognitive and noncognitive skills are modeled as low-dimensional latent variables. Estimates of these models show clear evidence of self- productivity and complementarity. Self-productivity becomes more evident as the child gets older. For cognitive skills, the complementarity between investments and skills tends to increase over the life cycle. As a result, compensating measures to offset adverse environments are less effective at later ages. Complementarity between noncognitive skills and investments displays a different pattern: The estimated effects are similar at different stages of child development, making it eas- ier to intervene with later remediation. Investments are particularly productive for disadvantaged children. The recognition of disadvantaged children as the optimal target for interventions also stands out in the recent study of Agostinelli & Wiswall (2016).17 Another focus of the literature is the importance of household and family characteristics in shaping child development. Many papers, such as those of Duncan et al. (1998), Akee et al. (2010), Bernal & Keane (2011), Dahl & Lochner (2012), Løken et al. (2012), and Agostinelli & Sorrenti (2018), study the impact of a family’s financial constraints on child development. Del Boca et al. (2014) estimate a production technology of child cognitive ability within an explicit model of household choices. Time inputs (both maternal and paternal) are found to be more important than monetary expenses in the child development process. Financial constraints matter more (and monetary transfers to a family are more effective) toward the end of the development process, during adolescence. Attanasio et al. (2018) exploit a randomized early childhood intervention in Colombia to estimate the technology of skill acquisition. Consistent with Del Boca et al. (2014), they find that the program led to positive effects on cognitive and socioemotional development primarily by inducing parents to increase their own investment in child development.18 How does the literature on skill formation relate to the economics of parenting styles? Tostart, we can view parenting style as yet another input in the production of skills by distinguishing different kinds of interactions between parents and children. A bigger difference is one of perspective.

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org The literature on skill formation, for the most part, limits itself to estimating a production tech- Access provided by Northwestern University on 10/02/19. For personal use only. nology, such as Equations 6 and 7 in the model used in this review. The theory of parenting style outlined in Sections 2 and 3, in contrast, formulates a full choice problem in which both parents and children make optimal decisions given the environment they face. Whereas the skill formation literature primarily asks which skills children will end up with for given inputs, the analysis in Section 3 is about explaining why parents and children make certain investment decisions. In the parenting style model, a major part of the impact of parents occurs not through direct in-

vestment (the parental inputs I1 and I2 in Equations 6 and 7), but instead through influencing the child’s own investment and choices (x1 and x2) through authoritarian or authoritative strategies.

17Other important contributions to the estimation of the technology of skill formation include those of Todd & Wolpin (2003, 2007), who model and estimate a dynamic child quality production function for cognitive skills to show the importance of both lagged and contemporaneous inputs on the production of current skills. 18Attanasio et al. (2017) provide a similar model for cognition and health of children aged 0–12 in India.

www.annualreviews.org • The Economics of Parenting 73 EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

Combining the two approaches therefore results in a broader perspective on the role of parents in child development. Conversely, the insights of the literature on skill formation are relevant for formulating better choice models of parenting style. The empirical analysis above is limited to broad classifications of types of parenting. In reality, there are many dimensions to parenting strategies, and parenting evolves over time as the child develops. For instance, the balance between coercion and persuasion in influencing children usually varies over time, depending on what is at stake and how thechild’s skills and abilities evolve. A research approach that combines the findings of the recent literature on skill formation with the choice-based approach of the literature on parenting styles should be productive for providing a deeper understanding of such issues. An explicit model of optimizing behavior of parents and children is also helpful for characterizing how decisions respond to public inputs; for example, in places where excellent public schools exist, and schools do not rely on additional investment at home (e.g., there is little homework), parents can relax more compared to places without such features. Lastly, the environment itself is also, to a large extent, a parental choice. Parents can decide in which neighborhood to live, with whom to socialize, and which schools their children should attend. This leads to a new set of interactions, which we explore in the next section.

5. THE INTERACTION OF PARENTING AND NEIGHBORHOODS IN CHILD DEVELOPMENT In the past decades, an extensive and multidisciplinary literature has studied the importance of neighborhoods in shaping an individual’s life opportunities (for reviews of neighborhood effects, see Jencks & Mayer 1990, Sampson et al. 2002). Most of the research focuses on children, especially children living in poor and distressed areas. Neighborhoods play an important role in part because, especially for older children, much of human capital accumulation takes place in schools, which may vary in quality and organization. Broadly defined neighborhoods also matter because they define the set of labor market opportunities that children face. Most young people start working in the same commuting zone in which they grew up. This affects the incentives for parents to adopt different parenting styles and to invest in their skill formation. More generally, the developmental psychology literature has long pointed out the importance of peers for child development. This is especially true for older children during adolescence, when attachment to parents is loosened, and relationships with peers take center stage. These observations do not imply that parents no longer matter once a child enters adolescence; rather, the way

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org in which parents influence their children shifts gradually from direct interactions to the parents’ Access provided by Northwestern University on 10/02/19. For personal use only. influence on the environment in which children live. For this reason, an encompassing theory of the economics of parenting should take into account how parents help determine the peers with whom their children interact, the schools that they attend, and the neighborhoods in which they live. We first discuss these issues from an empirical perspective, focusing on neighborhood effects in child development. We then outline how a choice of the child’s environment can be incorporated into more general models of the economics of parenting.

5.1. Empirical Evidence on Neighborhood Effects in Child Development Children who grow up in distressed areas tend to reach lower outcomes and display less upward mobility when compared to children from wealthier areas (Brooks-Gunn et al. 1993, Cutler & Glaeser 1997, Chetty et al. 2014). This stylized fact makes the analysis of neighborhood and peer effects a key point of theories of social capital accumulation (Coleman 1988) and human

74 Doepke • Sorrenti • Zilibotti EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

capital externalities and growth (Benabou 1993, Lucas 1988).19 One potential explanation underlying neighborhoods effects (Wilson 1987, Akerlof 1997, Glaeser & Scheinkman 2001) is that children in better neighborhoods are exposed to adults acting as role models who shape aspira- tions and adherence to social norms. Social networks and peer effects also play a role. For example, higher-income peers may have more information about labor market opportunities to be shared with the surrounding children (once grown up) and their parents (Manski 2000, Durlauf 1996). Neighborhoods may also affect the human capital accumulation process through community resources such as schools or libraries ( Jencks & Mayer 1990). The empirical study of neighborhood effects is a challenging task. If people’s location choices are driven in part by unobservable characteristics, then measured neighborhood effects based on observational data do not have a causal interpretation. One way to overcome this issue is to look at policies and interventions that induce experimental or quasi-experimental relocation of families. Perhaps the most studied intervention in this context is the Moving to Opportunity (MTO) experiment of the US Department of Housing and Urban Development. In the mid-1990s, the MTO program offered vouchers to a randomly selected group of families living in high-poverty housing projects that allowed them to relocate to areas with lower poverty rates. The experiment took place in the cities of Baltimore, Boston, Chicago, Los Angeles, and New York. Relocation generated large variation in neighborhood composition for otherwise similar families.20 Katz et al. (2001) analyze the short-run impacts of the MTO in the city of Boston. While individual economic self-sufficiency was unaffected by the displacement of families from higher- to lower-poverty areas, treated families experienced improvements in quality of life factors such as health, safety, and children’s behavioral outcomes. Kling et al. (2007) extend the analysis to all five cities involved in the MTO experiment and confirm the absence of effects on economic self- sufficiency.21 However, unlike in the Boston-only study, adult physical health did not improve in response to the intervention. Other outcomes for youths (such as health and risky behaviors) were differentially affected by gender. Heterogeneous effects of the MTO experiment also arise in the analysis of youth crime and delinquency by Kling et al. (2005). Ludwig et al. (2012, 2013) focus on longer-term outcomes and show that the MTO experiment positively affected adult mental and physical health and subjective well-being, whereas the impact on economic self-sufficiency or child achievement in education was negligible. One explanation for the limited measured impact of the MTO experiment is that not only neighborhood characteristics matter, but also when and how long one has lived there. Chetty et al. (2016) study the duration of exposure to a new neighborhood as a possible determinant of long-term effects on children.22 They find that (a) moving to a lower-poverty area improves long-

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org term outcomes of those children who moved at young ages, (b) the positive effect induced by the Access provided by Northwestern University on 10/02/19. For personal use only. move tends to decline with a child’s age at the time of the move, and (c) individuals moving during adolescence report negative outcomes induced by the move. Specifically, children moving to less

19For more information regarding the importance of peer effects, social interactions, and neighborhood effects, the reader is also referred to Case & Katz (1991), Borjas (1995), and Glaeser et al. (1996). Econometric issues that arise when there are unobserved neighborhood and individual characteristics in the presence of sorting are addressed by Altonji & Mansfield (2018) and Agrawal et al. (2019). 20Sanbonmatsu et al. (2011) provide a comprehensive description of the MTO experiment. 21These results are consistent with those of Oreopoulos (2003), who exploits differences in neighborhood quality of housing projects in Toronto to show little role for neighborhood effects in determining youths’ long-term earnings, unemployment status, and dependence on the welfare system. 22Long-term outcomes of disadvantaged children relocated to lower-poverty areas are also the object of study of Chyn (2018). By exploiting public housing demolition in Chicago during the 1990s, Chyn shows that moving out of disadvantaged neighborhoods increased a child’s employment and earnings in adulthood and re- duced arrests for violent crimes.

www.annualreviews.org • The Economics of Parenting 75 EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

poor areas before the age of 13 considerably increase their earnings in the mid-20s, are more likely to attend a college (and also a better college), are less likely to live in high-poverty areas, and (for women) are less likely to be single mothers. Conversely, moving when the child is between ages 13 and 18 has small or even negative effects on a child’s future outcomes. This finding suggests that the cost of moving (e.g., disruption of social networks and friendships) weighs more heavily once a child has reached adolescence. The importance of childhood exposure to neighborhoods is further investigated in recent papers studying tax records of millions of US families moving across different areas (Chetty & Hendren 2018a,b). Motivated by evidence that a child’s income conditional on parents’ income varies substantially with the area in which the child grew up (Chetty et al. 2014), Chetty & Hendren (2018a) exploit variation in the age of children when a family moves from one commuting zone to another to estimate the impact of neighborhoods on intergenerational mobility.23 They find that, “on average, spending an additional year in a commuting zone where themean income rank of children of permanent residents is 1 percentile higher (at a given level of parental income) increases a child’s income rank in adulthood by approximately 0.04 percentiles” (Chetty & Hendren 2018a, p. 1110).24 Chetty & Hendren (2018b) examine the causal effect of each county in the United States on children’s outcomes in adulthood to provide insights about the mechanism underlying neighborhood effects. Previous analyses by Sampson et al. (2002) and Chetty et al. (2014) show correlations between upward mobility and factors such as residential segregation, income inequality, demographic characteristics (fraction of single mothers and racial composition), and the availabil- ity of good primary schools. Although these correlations are informative, they fail to disentangle the causal effect of a specific neighborhood from endogenous sorting. The evidence of Chetty& Hendren (2018b) suggests that the association between upward mobility and group-level characteristics such as racial and income segregation or income inequality is mostly driven by the direct effect of neighborhoods, whereas the relationship between upward mobility and average individual-level characteristics (such as family structure) is due to sorting. Summing up, the literature suggests that neighborhoods are important determinants of a child’s future opportunities. Understanding the mechanisms underlying neighborhood effects remains a research priority.25

5.2. Neighborhood Choice in a Model of Parenting From the perspective of the economics of parenting, the results of this literature suggest that par-

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org ents have a major impact on their children through determining their environment, including (but Access provided by Northwestern University on 10/02/19. For personal use only. not limited to) the choice of which neighborhood to live in. To date, there is little work from a modeling perspective on neighborhood choice as an aspect of parenting, and none that considers

23The causal interpretation of their results relies on the assumption that the selection effect (e.g., a child’s unobservable characteristics) associated with a move to a better neighborhood is invariant with respect to the child’s age at the time of the move. 24Interestingly, while in the Chetty et al. (2016) study of the MTO experiment only young children benefit from a better neighborhood, Chetty & Hendren (2018a) find that the neighborhood effect persists until an individual reaches their early 20s. The difference can be rationalized as a consequence of the two different settings: In the MTO experiment, movers are compared to nonmovers, while Chetty & Hendren (2018a) compare movers across different commuting zones, which nets out fixed disruption costs of moving. 25Another recent contribution to this issue is that of Gibbons et al. (2013, 2017), who study the determinants of students’ educational outcomes in England using a model with neighborhood fixed effects. Neighborhood composition is found to play only a limited role in teenagers’ test scores, whereas turnover of same-school- grade students reduces value added for students who stay in their original neighborhood.

76 Doepke • Sorrenti • Zilibotti EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

how this important dimension of parenting interacts with other parenting choices, such as that of a parenting style. Eckert & Kleineberg (2019) estimate a model of neighborhood choice where the value of the neighborhood is exogenous (and estimated). In the work of Fogli & Guerrieri (2018), endogenous neighborhood effects arise from a human capital spillover. Agostinelli (2018) estimates a dynamic model of skill formation where children choose their own peer groups, and parental investments respond to the children’s peer groups (although the choice of neighborhood is not modeled). In ongoing work, Agostinelli et al. (2019) develop a model of endogenous neighborhood choice and endogenous neighborhood characteristics within a theory of parenting along the lines of the model described in Section 2. In the work of Agostinelli et al. (2019), there are explicit peer effects, and there is endogenous sorting due to endogenous rental costs. Tooutline the main ideas, consider a version of the model described in Section 2 in which there are two neighborhoods, P (posh) and B (bohemian). Thus,

the neighborhood variable dt that enters the technology of skill formation in Equations 6 and 7 ∈{ } satisfies dt P, B .Rentalpricesq(dt ) that enter the budget constraint in Equation 5 may differ across these neighborhoods. The new element is that neighborhood choice and neighborhood quality are endogenous. In choosing a neighborhood, parents consider their own preferences, the quality of the neighborhood as a place to raise children, and the cost of living in the neighborhood. Parents’ own preferences are determined by a random preference shock for neighborhoods. For instance, parents may have a preference for the neighborhood in which they grew up. The quality of the neighborhood is driven by peer effects. As a simple example, suppose that the quality of the neighborhood depends on the average skills S of the adult residents. In more elaborate formulations, neighborhood quality can also depend on local public goods; parents’ investments I; and the skills s and educational effort x of the local children, that is, peer effects. The model implies that both parenting and neighborhood choice interact with economic conditions. The model displays a force toward economic segregation: Richer parents are more willing and more able to pay high rents to provide their children with a good environment. In the basic model, the strength of this force depends primarily on inequality. Consider first an environment w /w with low inequality, which could be modeled as a low ratio H L in the setup used in Section 3.2. The equilibrium will display some segregation, but only to a limited extent. In particular, there will be one neighborhood where rich people are more likely to live, and another that has more poor people. The poorer households are deterred from the rich neighborhood by higher rental costs. Nevertheless, with low inequality, the degree of sorting will also be low. If inequality is low, then so are the stakes of getting the best possible education for the children. Thus, providing the best

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org possible environment for the children is a relatively weak force in choosing the neighborhood, and Access provided by Northwestern University on 10/02/19. For personal use only. for many parents, their preference shock (which is uncorrelated with status) will drive the choice. The difference in the rental price across the neighborhoods will also be low, again because parents have a limited willingness to pay for a better environment for their children. Now consider the same economy but with higher inequality. Parents will now be more concerned about their children’s skill accumulation, and consequently, the willingness to pay for a better neighborhood goes up. The result is more segregation: There will be less mixing between rich and poor people, and the difference in the rental cost between the neighborhoods will increase. Notice that, while in this case we consider the underlying inequality as an exogenous force, spatial sorting has a self-reinforcing element. As the rich and poor become more separated, the difference in the educational environment in the two neighborhoods (which, recall, we relate to the average skill level S of the local population) will also increase. This further increases the willingness of the rich to pay for living in the better neighborhood.

www.annualreviews.org • The Economics of Parenting 77 EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

Another self-reinforcing mechanism works through parents’ investments in their children’s skill accumulation. If neighborhood quality and parental investments are complementary, then increased segregation leads to even stronger parenting investments among the rich and less investment among the poor in the disadvantaged neighborhood. Such complementarities can arise from a variety of sources. For example, when children have many peers who are strongly focused on educational achievement, they may be more receptive to their parents’ efforts compared to children in an environment where the peer group shows little interest in school. There are also interactions with the choice of parenting style. For instance, in poor and crime-ridden neighborhoods, parents are likely to emphasize risk aversion, since many of the risks that are present in the environment (such as juvenile crime and drugs) can have harmful long-term consequences. A side effect is that the children may end up so risk averse that they are less inclined to jump on economic opportunities that involve some risk taking. In contrast, in safer middle-class environments, parents can afford to be more permissive and risk tolerant, which translates into a more positive attitude of children toward opportunities that present some upside risk. This discussion shows that even a relatively simple model of neighborhood choice and parenting based on the simple framework presented in Section 2 can address many of the trends over recent decades that we discuss above, such as the increase in economic inequality, rising residential segregation, and a rising gap in parental investments between families from different ends of the social spectrum. A better understanding of the mutual interactions that give rise to these trends will be helpful in formulating policy responses that can be used to prevent these trends from accelerating further.

6. CONCLUSIONS AND DIRECTIONS FOR FUTURE RESEARCH The economics of parenting has come a long way since the family first entered the realm of economics with the work of Gary Becker. Economic models turn out to be remarkably successful at explaining many dimensions of what parents do, ranging from fertility choice (which we do not discuss in detail in this review) and parenting style to the choice of a neighborhood for the family to live in. There has also been tremendous progress on understanding the technology of children’s skill formation. The findings of this literature already play a crucial role in policy making; forex- ample, the observation that many important skills are acquired during the first years of life has led to calls for more support for high-quality daycare and preschool, which is now being imple- mented in many places. As this example shows, the economics of parenting takes a central place in some major policy challenges of this age. Rising inequality; increased segregation; low social

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org mobility; and, more generally, increased polarization in society are trends that most industrialized Access provided by Northwestern University on 10/02/19. For personal use only. countries confront today. Further research on the economics of parenting will help advance our understanding of the root causes of these trends and can clarify options for public policy to address the resulting challenges. There are many remaining research challenges for the economics of parenting. One example is the measurement of parenting styles. Existing measures, which derive from developmental psychology, only provide a broad classification, and different approaches can be incompatible with each other. It would be fruitful to combine detailed time-use data with other information on the relationships of parents and children to construct more finely graded measures that map more directly into economic models of parenting. A broader challenge for future work is to combine the three strands of research on the economics of parenting discussed in this review. To date, the economics of parenting style, the technology of skill formation, and the role of neighborhood effects have been mostly studied in isolation. However, these dimensions are clearly closely related: Parenting style is a crucial input in

78 Doepke • Sorrenti • Zilibotti EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

skill formation, and so is the choice of a neighborhood. Conversely, the neighborhood in which a family lives and thus the risks and opportunities that a child is exposed to matter for the choice of a parenting style. One recent paper that combines two of these elements is that of Deckers et al. (2017), who relate socioeconomic differences among children in IQ, patience, and other economic preferences to parental and family characteristics, parental time investments, and parenting style and show that parenting style and time investments play distinct roles. There is a lot of scope for future work that takes interactions between the different aspects of the economics of parenting into account. Another limitation of existing research is that most theoretical models of parent–child interaction focus on the case of a single parent and a single child. In reality, there are two biological parents, but many children live with a single parent, and step-parents are also present in many families. It would be fruitful to take more account of interactions among multiple parents and caregivers in economic models of parenting. Likewise, in many families, there are multiple children, and we know from empirical work that birth order is an important determinant of children’s outcomes (Black et al. 2005). For parents, it quickly becomes apparent that the choice of parenting style for, say, the third child is more constrained than for the first, given the competing needs of the older children (see Hao et al. 2008). Children also come with their own personalities (and genders), and parents use different strategies with different children in response to their characteristics. A lot of productive research could be done extending theoretical models of parenting beyond the single-parent, single-child case to capture the multifaceted interactions among many family members.26 Beyond refining the economics of parenting, more research is also needed to assess theimpli- cations of changes in parenting for broader economic issues. For example, if parenting gets more time intensive, then we might expect fertility rates to fall. The general relationship between investment in children and fertility is a major topic in the economics of fertility (i.e., the quantity–quality trade-off ), but we are not aware of work that makes a direct connection between a rising intensity of parenting and the currently very low fertility rates in many industrialized countries (Doepke & Kindermann 2019), that reconsiders the income–fertility gradient from the perspective of parenting decisions, or that formulates a joint decision problem involving fertility and parenting style. Parenting also interacts with labor supply and careers, especially for mothers (Adda et al. 2017). Recent research on the gender wage gap has identified motherhood as the primary reason why women still get paid less than men (Gallen 2018, Kleven et al. 2018). For the impact of parent- hood on careers, choices such as parenting style and time spent on parenting are surely important, which is another connection not yet explored in the literature.

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org Lastly, given the central role of parenting in the accumulation of skills and human capital, the Access provided by Northwestern University on 10/02/19. For personal use only. economics of parenting is highly relevant for macroeconomic issues such as the dynamics of inequality, intergenerational mobility, and overall economic growth. There are only a few papers that take recent advances in research on the technology of skill formation into account while developing multigeneration models to study determinants of the intergenerational transmission of economic status (Yum 2015, Caucutt & Lochner 2017, Lee & Seshadri 2019). We are aware of only a single paper, namely that of Daruich (2018), that examines the importance of early investments in children, a major theme in literature on the technology of skill formation, for the evolution of inequality and growth in a dynamic macroeconomic model. In the future, we expect that macroeconomic modeling will be one of the areas to particularly benefit from ongoing research on the economics of parenting.

26Two recent examples of work along these lines are the work of Blandin & Herrington (2018), who consider the impact of family structure and early family investments on college completion rates, and Gayle et al. (2015), who analyze racial differences in human capital accumulation.

www.annualreviews.org • The Economics of Parenting 79 EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

DISCLOSURE STATEMENT The authors are not aware of any affiliations, memberships, funding, or financial holdings that might be perceived as affecting the objectivity of this review.

ACKNOWLEDGMENTS We thank Ashley Wong and Jane Olmstead-Rumsey for excellent research assistance. Financial support from the National Science Foundation (grant SES-1260961) and the Swiss National Sci- ence Foundation (grant 100018_165616 to G.S. and F.Z.) is gratefully acknowledged.

LITERATURE CITED Adda J, Dustmann C, Stevens K. 2017. The career costs of children. J. Political Econ. 125(2):293–337 Agostinelli F. 2018. Investing in children’s skills: an equilibrium analysis of social interactions and parental investments. Unpublished manuscript, Ariz. State Univ., Tempe Agostinelli F, Doepke M, Sorrenti G, Zilibotti F. 2019. It takes a village: the economics of parenting with neighborhood and peer effects. Unpublished manuscript, Yale Univ., New Haven, CT Agostinelli F, Sorrenti G. 2018. Money versus time: family income, maternal labor supply, and child development. Work. Pap. 2018-017, Hum. Cap. Econ. Oppor. Glob. Work. Group, Univ. Chicago Agostinelli F, Wiswall M. 2016. Estimating the technology of children’s skill formation. NBER Work. Pap. 22442 Agrawal M, Altonji JG, Mansfield RK. 2019. Quantifying family, school, and location effects in the presence of complementarities and sorting. J. Labor Econ. 37(S1):S11–83 Akee RKQ, Copeland WE, Keeler G, Angold A, Costello EJ. 2010. Parents’ incomes and children’s outcomes: a quasi-experiment using transfer payments from casino profits. Am. Econ. J. Appl. Econ. 2(1):86–115 Akerlof GA. 1997. Social distance and social decisions. Econometrica 65(5):1005–27 Almlund M, Duckworth A, Heckman JJ, Kautz T. 2011. Personality psychology and economics. In Handbook of the Economics of Education, Vol. 4, ed. EA Hanushek, S Machin, L Woessmann, pp. 1–181. Amsterdam: Elsevier Altonji JG, Mansfield RK. 2018. Estimating group effects using averages of observables to control forsorting on unobservables: school and neighborhood effects. Am. Econ. Rev. 108(10):2902–46 Andrew Young School Policy Stud. 2010. Andrew Young School World Tax Indicators. Atlanta, GA: Int. Stud. Progr. Attanasio O. 2015. The determinants of human capital formation during the early years of life: theory, measurement, and policies. J. Eur. Econ. Assoc. 13(6):949–97 Attanasio O, Cattan S, Fitzsimmons E, Meghir C, Rubio-Codina M. 2018. Estimating the production function for

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org human capital: results from a randomized controlled trial in Colombia. Work. Pap. W18/18, Inst. Fisc. Stud., Access provided by Northwestern University on 10/02/19. For personal use only. London Attanasio O, Meghir C, Nix E. 2017. Human capital development and parental investment in India. NBER Work. Pap. 21740 Bandura A. 1986. Social Foundations of Thought and Actions. Upper Saddle River, NJ: Prentice Hall Barro RJ, Becker GS. 1989. Fertility choice in a model of economic growth. Econometrica 57(2):481–501 Bartkowski JP, Ellison CG. 1995. Divergent models of childrearing in popular manuals: conservative Protes- tants versus the mainstream experts. Sociol. Relig. 56(1):21–34 Baumrind D. 1967. Child care practices anteceding three patterns of preschool behavior. Genet. Psychol. Monogr. 75(1):43–88 Becker GS. 1960. An economic analysis of fertility. In Demographic and Economic Change in Developed Countries, ed. GB Roberts, pp. 209–40. Princeton, NJ: Princeton Univ. Press Becker GS. 1981. A Treatise on the Family. Cambridge, MA: Harvard Univ. Press Becker GS, Barro RJ. 1988. A reformulation of the economic theory of fertility. Q. J. Econ. 103(1):1–25 Becker GS, Mulligan CB. 1997. The endogenous determination of time preference. Q. J. Econ. 112(3):729–58

80 Doepke • Sorrenti • Zilibotti EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

Becker GS, Murphy KM, Spenkuch JL. 2016. The manipulation of children’s preferences, old age support, and investment in children’s human capital. J. Labor Econ. 34(S2):S3–30 Becker GS, Tomes N. 1979. An equilibrium theory of the distribution of income and intergenerational mobility. J. Political Econ. 87(6):1153–89 Becker GS, Tomes N. 1986. Human capital and the rise and fall of families. J. Labor Econ. 4(3):S1–39 Benabou R. 1993. Workings of a city: location, education, and production. Q. J. Econ. 108(3):619–52 Bernal R, Keane MP. 2011. Child care choices and children’s cognitive achievement: the case of single mothers. J. Labor Econ. 29(3):459–512 Bisin A, Verdier T. 2001. The economics of cultural transmission and the dynamics of preferences. J. Econ. Theory 97(2):298–319 Bisin A, Verdier T. 2011. The economics of cultural transmission and socialization. In Handbook of Social Economics, Vol. 1A, ed. J Benhabib, A Bisin, MO Jackson, pp. 339–416. Amsterdam: Elsevier Black SE, Devereux PJ, Salvanes KG. 2005. The more the merrier? The effect of family size and birth order on children’s education. Q. J. Econ. 120(2):669–700 Blandin A, Herrington C. 2018. Family structure, human capital investment, and aggregate college attainment.Un- published manuscript, Va. Commonw. Univ., Richmond Borjas GJ. 1995. Ethnicity, neighborhoods, and human-capital externalities. Am. Econ. Rev. 85(3):365–90 Brooks-Gunn J, Duncan GJ, Klebanov PK, Sealand N. 1993. Do neighborhoods influence child and adolescent development? Am. J. Sociol. 99(2):353–95 Browning M, Chiappori P-A, Weiss Y. 2014. Economics of the Family. Cambridge, UK: Cambridge Univ. Press Burton P, Phipps S, Curtis L. 2002. All in the family: a simultaneous model of parenting style and child conduct. Am. Econ. Rev. 92(2):368–72 Carneiro P, Heckman JJ. 2003. Human capital policy. In Inequality in America: What Role for Human Capital Policies?, ed. JJ Heckman, AB Krueger, pp. 77–239. Cambridge, MA: MIT Press Case A, Katz LF. 1991. The company you keep: the effects of family and neighborhood on disadvantaged youths.NBER Work. Pap. 3705 Caucutt EM, Lochner L. 2017. Early and late human capital investments, borrowing constraints, and the family. Work. Pap. 2017-3, Cent. Hum. Cap. Product., Univ. West. Ont., London, Can. Cawley J, Heckman JJ, Vytlacil EJ. 2001. Three observations on wages and measured cognitive ability. Labor Econ. 8(4):419–42 Chetty R, Hendren N. 2018a. The impacts of neighborhoods on intergenerational mobility I: childhood exposure effects. Q. J. Econ. 133(3):1107–62 Chetty R, Hendren N. 2018b. The impacts of neighborhoods on intergenerational mobility II: county-level estimates. Q. J. Econ. 133(3):1163–228 Chetty R, Hendren N, Katz LF. 2016. The effects of exposure to better neighborhoods on children: new evidence from the moving to opportunity experiment. Am. Econ. Rev. 106(4):855–902 Chetty R, Hendren N, Kline P, Saez E. 2014. Where is the land of opportunity? The geography of intergen-

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org erational mobility in the United States. Q. J. Econ. 129(4):1553–623

Access provided by Northwestern University on 10/02/19. For personal use only. Chiappori P-A, Salanié B, Weiss Y. 2017. Partner choice, investment in children, and the marital college premium. Am. Econ. Rev. 107(8):2109–67 Chyn E. 2018. Moved to opportunity: the long-run effects of public housing demolition on children. Am. Econ. Rev. 108(10):3028–56 Cobb-Clark DA, Salamanca N, Zhu A. 2019. Parenting style as an investment in human development. J. Popul. Econ. In press Coleman JS. 1988. Social capital in the creation of human capital. Am. J. Sociol. 94:S95–120 Cosconati M. 2013. Optimal parenting styles: evidence from a dynamic game with multiple equilibria. Unpublished manuscript, Bank Italy, Rome Cunha F. 2015. Subjective rationality, parenting styles, and investments in children. In Families in an Era of Increasing Inequality: Diverging Destinies, ed. PR Amato, A Booth, SM McHale, J Van Hook, pp. 83–94. Berlin: Springer Cunha F, Heckman JJ. 2007. The technology of skill formation. Am. Econ. Rev. 97(2):31–47

www.annualreviews.org • The Economics of Parenting 81 EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

Cunha F, Heckman JJ. 2008. Formulating, identifying, and estimating the technology for the formation of skills. J. Hum. Resour. 43(4):738–82 Cunha F, Heckman JJ, Lochner L, Masterov DV. 2006. Interpreting the evidence on life cycle skill formation. In Handbook of the Economics of Education, Vol. 1, ed. EA Hanushek, F Welch, pp. 697–812. Amsterdam: Elsevier Cunha F, Heckman JJ, Schennach SM. 2010. Estimating the technology of cognitive and noncognitive skill formation. Econometrica 78(3):883–931 Cutler DM, Glaeser EL. 1997. Are ghettos good or bad? Q. J. Econ. 112(3):827–72 Dahl GB, Lochner L. 2012. The impact of family income on child achievement: evidence from the Earned Income Tax Credit. Am. Econ. Rev. 102(5):1927–56 Daruich D. 2018. The macroeconomic consequences of early childhood development policies. Unpublished manuscript, New York Univ. Deckers T, Falk A, Kosse F, Pinger P, Schildberg-Hörisch H. 2017. Socio-economic status and inequalities in children’s IQ and economic preferences. Discuss. Pap. 11158, Inst. Labor Econ., Bonn, Ger. Del Boca D, Flinn C, Wiswall M. 2014. Household choices and child development. Rev. Econ. Stud. 81(1):137– 85 Doepke M. 2015. Gary Becker on the quantity and quality of children. J. Demogr. Econ. 81(1):59–66 Doepke M, Kindermann F. 2019. Bargaining over babies: theory, evidence, and policy implications. Am. Econ. Rev.Inpress Doepke M, Tertilt M. 2016. Families in macroeconomics. In Handbook of Macroeconomics, Vol. 2, ed. JB Taylor, H Uhlig, pp. 1789–891. Amsterdam: Elsevier Doepke M, Zilibotti F. 2005. Social class and the spirit of capitalism. J. Eur. Econ. Assoc. 3(2–3):516–24 Doepke M, Zilibotti F. 2008. Occupational choice and the spirit of capitalism. Q. J. Econ. 123(2):747–93 Doepke M, Zilibotti F. 2017. Parenting with style: altruism and paternalism in intergenerational preference transmission. Econometrica 85(5):1331–71 Doepke M, Zilibotti F. 2019. Love, Money, and Parenting: How Economics Explains the Way We Raise Our Kids. Princeton, NJ: Princeton Univ. Press Dohmen T, Enke B, Falk A, Huffman D, Sunde U. 2018. Patience and contemporary development. Unpublished manuscript, Univ. Bonn, Ger. Dohmen T, Falk A, Huffman D, Sunde U. 2012. The intergenerational transmission of risk and trust attitudes. Rev. Econ. Stud. 79(2):645–77 Duncan GJ, Yeung WJ, Brooks-Gunn J, Smith JR. 1998. How much does childhood poverty affect the life chances of children? Am. Sociol. Rev. 63(3):406–23 Durlauf SN. 1996. A theory of persistent income inequality. J. Econ. Growth 1(1):75–93 Eckert F, Kleineberg T. 2019. Can we save the American dream? A dynamic general equilibrium analysis of the effects of school financing on local opportunities. Unpublished manuscript, Yale Univ., New Haven, CT Erikson EH. 1950. Childhood and Society. New York: Norton

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org Ermisch J, Francesconi M. 2005. Parental employment and children’s welfare. In Women at Work: An Economic

Access provided by Northwestern University on 10/02/19. For personal use only. Perspective, ed. T Boeri, D Del Boca, C Pissarides, pp. 154–93. Oxford, UK: Oxford Univ. Press Figlio D, Giuliano P, Özek U, Sapienza P. 2016. Long-term orientation and educational performance.NBER Work. Pap. 22541 Fogli A, Guerrieri V. 2018. The end of the American dream? Inequality and segregation in US cities. Unpublished manuscript, Univ. Chicago Gallen Y. 2018. Motherhood and the gender productivity gap. Unpublished manuscript, Univ. Chicago Gayle G-L, Golan L, Soytas MA. 2015. What accounts for the racial gap in time allocation and intergenerational transmission of human capital? Work. Pap. 2015–18, Fed. Reserve Bank St. Louis, MO Gibbons S, Silva O, Weinhardt F. 2013. Everybody needs good neighbours? Evidence from students’ outcomes in England. Econ. J. 123(571):831–74 Gibbons S, Silva O, Weinhardt F. 2017. Neighbourhood turnover and teenage attainment. J. Eur. Econ. Assoc. 15(4):746–83 Glaeser E, Scheinkman J. 2001. Measuring social interactions. In Social Dynamics, ed. S Durlauf, P Young, pp. 83–130. Cambridge, MA: MIT Press

82 Doepke • Sorrenti • Zilibotti EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

Glaeser EL, Sacerdote B, Scheinkman JA. 1996. Crime and social interactions. Q. J. Econ. 111(2):507–48 Greenwood J, Guner N, Vandenbroucke G. 2017. Family economics writ large. J. Econ. Lit. 55(4):1346–434 Hao L, Hotz VJ, Jin GZ. 2008. Games parents and adolescents play: risky behaviour, parental reputation and strategic transfers. Econ. J. 118(528):515–55 Heckman JJ, Masterov DV. 2007. The productivity argument for investing in young children. Rev. Agric. Econ. 29(3):446–93 Heckman JJ, Mosso S. 2014. The economics of human development and social mobility. Annu. Rev. Econ. 6:689–733 Heckman JJ, Pinto R, Savelyev PA. 2013. Understanding the mechanisms through which an influential early childhood program boosted adult outcomes. Am. Econ. Rev. 103(6):2052–86 Heckman JJ, Rubinstein Y. 2001. The importance of noncognitive skills: lessons from the GED testing program. Am. Econ. Rev. 91(2):145–49 Heckman JJ, Stixrud J, Urzua S. 2006. The effects of cognitive and noncognitive abilities on labor market outcomes and social behavior. J. Labor Econ. 24(3):411–82 Hotz VJ, Klerman JA, Willis RJ. 1997. The economics of fertility in developed countries. In Handbook of Population and Family Economics, Vol. 1A, ed. MR Rosenzweig, O Stark, pp. 275–347. Amsterdam: Elsevier Jencks C, Mayer SE. 1990. The social consequences of growing up in a poor neighborhood. In Inner-City Poverty in the United States, ed. EL Laurence Jr., MGH McGeary, pp. 111–85. Washington, DC: Natl. Acad. Press Katz LF, Kling JR, Liebman JB. 2001. Moving to opportunity in Boston: early results of a randomized mobility experiment. Q. J. Econ. 116(2):607–54 Kleven H, Landais C, Sgaard JE. 2018. Children and gender inequality: evidence from Denmark. NBER Work. Pap. 24219 Kling JR, Liebman JB, Katz LF. 2007. Experimental analysis of neighborhood effects. Econometrica 75(1):83– 119 Kling JR, Ludwig J, Katz LF. 2005. Neighborhood effects on crime for female and male youth: evidence from a randomized housing voucher experiment. Q. J. Econ. 120(1):87–130 Knudsen EI, Heckman JJ, Cameron JL, Shonkoff JP. 2006. Economic, neurobiological, and behavioral per- spectives on building America’s future workforce. PNAS 103(27):10155–62 Lee SY, Seshadri A. 2019. On the intergenerational transmission of economic status. J. Political Econ. In press Lizzeri A, Siniscalchi M. 2008. Parental guidance and supervised learning. Q. J. Econ. 123(3):1161–95 Løken KV, Mogstad M, Wiswall M. 2012. What linear estimators miss: the effects of family income on child outcomes. Am. Econ. J. Appl. Econ. 4(2):1–35 Lucas RE. 1988. On the mechanics of economic development. J. Monet. Econ. 22(1):3–42 Ludwig J, Duncan GJ, Gennetian LA, Katz LF, Kessler RC, et al. 2012. Neighborhood effects on the long- term well-being of low-income adults. Science 337(6101):1505–10 Ludwig J, Duncan GJ, Gennetian LA, Katz LF, Kessler RC, et al. 2013. Long-term neighborhood effects on

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org low-income families: evidence from moving to opportunity. Am. Econ. Rev. 103(3):226–31

Access provided by Northwestern University on 10/02/19. For personal use only. Lundberg S, Pollak RA. 2007. The American family and family economics. J. Econ. Perspect. 21(2):3–26 Lundberg S, Pollak RA, Stearns J. 2016. Family inequality: diverging patterns in marriage, cohabitation, and childbearing. J. Econ. Perspect. 30(2):79–102 Lundberg S, Romich JL, Tsang KP. 2009. Decision-making by children. Rev. Econ. Househ. 7(1):1–30 Manski CF. 2000. Economic analysis of social interactions. J. Econ. Perspect. 14(3):115–36 Montenegro CE, Patrinos HA. 2014. Comparable estimates of returns to schooling around the world. Policy Res. Work. Pap. WPS 7020, World Bank, Washington, DC Murnane RJ, Willett JB, Levy F. 1995. The growing importance of cognitive skills in wage determination. Rev. Econ. Stat. 77(2):251–66 Oreopoulos P. 2003. The long-run consequences of living in a poor neighborhood. Q. J. Econ. 118(4):1533– 75 Ramey G, Ramey VA. 2010. The rug rat race. Brook. Pap. Econ. Act. 41:129–76 Romich JL, Lundberg S, Tsang KP. 2009. Independence giving or autonomy taking? Childhood predictors of decision-sharing patterns between young adolescents and parents. J. Res. Adolesc. 19(4):587–600

www.annualreviews.org • The Economics of Parenting 83 EC11CH03_Doepke ARjats.cls July 16, 2019 14:48

Saez Marti M. 2018. Observational learning and parental influence. Unpublished manuscript, Yale Univ., New Haven, CT Sampson RJ, Morenoff JD, Gannon-Rowley T. 2002. Assessing “neighborhood effects”: social processes and new directions in research. Annu. Rev. Sociol. 28:443–78 Sanbonmatsu L, Ludwig J, Katz LF, Gennetian LA, Duncan GJ, et al. 2011. Moving to Opportunity for Fair Housing Demonstration Program: Final Impacts Evaluation. Washington, DC: Off. Policy Dev. Res., US Dep. Hous. Urban Dev. Seror A. 2019. Intergenerational rearing and the transmission of a cultural cognition. Unpublished manuscript, Chapman Univ., Orange, CA Sutter M, Kocher MG, Glaetze-Ruetzler D, Trautmann ST. 2013. Impatience and uncertainty: experimental decisions predict adolescents’ field behavior. Am. Econ. Rev. 103(1):510–31 Todd PE, Wolpin KI. 2003. On the specification and estimation of the production function for cognitive achievement. Econ. J. 113(485):F3–33 Todd PE, Wolpin KI. 2007. The production of cognitive achievement in children: home, school, and racial test score gaps. J. Hum. Cap. 1(1):91–136 Weinberg BA. 2001. An incentive model of the effect of parental income on children. J. Political Econ. 109(2):266–80 Wilson WJ. 1987. The Truly Disadvantaged: The Inner City, the Underclass, and Public Policy. Chicago: Univ. Chicago Press Yum M. 2015. Parental time investment and intergenerational mobility. Unpublished manuscript, Univ. Mannheim, Ger. Zhang L, Ikeda S. 2016. Welfare-enhancing parental altruism and children’s habit formation. Int. Rev. Econ. 63(3):281–303 Zumbuehl M, Dohmen T, Pfann G. 2013. Parental investment and the intergenerational transmission of economic preferences and attitudes. Discuss. Pap. 7476, Inst. Labor Econ., Bonn, Ger. Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org Access provided by Northwestern University on 10/02/19. For personal use only.

84 Doepke • Sorrenti • Zilibotti EC11_FrontMatter ARI 29 July 2019 13:43

Annual Review of Economics Contents Volume 11, 2019

The Economics of Kenneth J. Arrow: A Selective Review Eric S. Maskin pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp1 Econometrics of Auctions and Nonlinear Pricing Isabelle Perrigne and Quang Vuong pppppppppppppppppppppppppppppppppppppppppppppppppppppppppp27 The Economics of Parenting Matthias Doepke, Giuseppe Sorrenti, and Fabrizio Zilibotti pppppppppppppppppppppppppppppppp55 Markets for Information: An Introduction Dirk Bergemann and Alessandro Bonatti pppppppppppppppppppppppppppppppppppppppppppppppppppp85 Global Wealth Inequality Gabriel Zucman ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp109 Robustness in Mechanism Design and Contracting Gabriel Carroll pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp139 Experiments on Cognition, Communication, Coordination, and Cooperation in Relationships Vincent P. Crawford pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp167 Bootstrap Methods in Econometrics Joel L. Horowitz ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp193 Experiments and Entrepreneurship in Developing Countries Simon Quinn and Christopher Woodruff pppppppppppppppppppppppppppppppppppppppppppppppppp225

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org Bayesian Persuasion and Information Design Access provided by Northwestern University on 10/02/19. For personal use only. Emir Kamenica pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp249 Transitional Dynamics in Aggregate Models of Innovative Investment Andrew Atkeson, Ariel T. Burstein, and Manolis Chatzikonstantinou pppppppppppppppppp273 Echo Chambers and Their Effects on Economic and Political Outcomes Gilat Levy and Ronny Razin pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp303 Evolutionary Models of Preference Formation Ingela Alger and J¨orgen W. Weibull ppppppppppppppppppppppppppppppppppppppppppppppppppppppp329 Approximately Optimal Mechanism Design Tim Roughgarden and Inbal Talgam-Cohen pppppppppppppppppppppppppppppppppppppppppppppp355

v EC11_FrontMatter ARI 29 July 2019 13:43

Auction Market Design: Recent Innovations Paul Milgrom pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp383 Fair Division in the Internet Age Herv´e Moulin pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp407 Legislative and Multilateral Bargaining H¨ulya Eraslan and Kirill S. Evdokimov ppppppppppppppppppppppppppppppppppppppppppppppppppp443 Social Networks in Policy Making Marco Battaglini and Eleonora Patacchini ppppppppppppppppppppppppppppppppppppppppppppppppp473 Econometric Analysis of Panel Data Models with Multifactor Error Structures Hande Karabiyik, Franz C. Palm, and Jean-Pierre Urbain ppppppppppppppppppppppppppppp495 Using Randomized Controlled Trials to Estimate Long-Run Impacts in Development Economics Adrien Bouguen, Yue Huang, Michael Kremer, and Edward Miguel ppppppppppppppppppp523 Is Education Consumption or Investment? Implications for School Competition W. Bentley MacLeod and Miguel Urquiola pppppppppppppppppppppppppppppppppppppppppppppppp563 Productivity Measurement: Racing to Keep Up Daniel E. Sichel ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp591 History, Microdata, and Endogenous Growth Ufuk Akcigit and Tom Nicholas pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp615 Production Networks: A Primer Vasco M. Carvalho and Alireza Tahbaz-Salehi ppppppppppppppppppppppppppppppppppppppppppp635 Economic Theories of Justice Marc Fleurbaey pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp665 Machine Learning Methods That Economists Should Know About pppppppppppppppppppppppppppppppppppppppppppppppppppppppp

Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org Susan Athey and Guido W. Imbens 685 Access provided by Northwestern University on 10/02/19. For personal use only. Weak Instruments in Instrumental Variables Regression: Theory and Practice Isaiah Andrews, James H. Stock, and Liyang Sun pppppppppppppppppppppppppppppppppppppppp727 Taking State-Capacity Research to the Field: Insights from Collaborations with Tax Authorities Dina Pomeranz and Jos´e Vila-Belda ppppppppppppppppppppppppppppppppppppppppppppppppppppppp755 Free Movement, Open Borders, and the Global Gains from Labor Mobility Christian Dustmann and Ian P. Preston ppppppppppppppppppppppppppppppppppppppppppppppppppp783

vi Contents EC11_FrontMatter ARI 29 July 2019 13:43

Monetary Policy, Macroprudential Policy, and Financial Stability David Martinez-Miera and Rafael Repullo ppppppppppppppppppppppppppppppppppppppppppppppp809 Has Dynamic Programming Improved Decision Making? John Rust pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp833 The International Monetary and Financial System Pierre-Olivier Gourinchas, H´el`ene Rey, and Maxime Sauzet pppppppppppppppppppppppppppp859

Symposium: Universal Basic Income

Universal Basic Income: Some Theoretical Aspects Maitreesh Ghatak and Fran¸cois Maniquet pppppppppppppppppppppppppppppppppppppppppppppppp895 Universal Basic Income in the United States and Advanced Countries Hilary Hoynes and Jesse Rothstein pppppppppppppppppppppppppppppppppppppppppppppppppppppppppp929 Universal Basic Income in the Developing World Abhijit Banerjee, Paul Niehaus, and Tavneet Suri pppppppppppppppppppppppppppppppppppppppp959

Indexes

Cumulative Index of Contributing Authors, Volumes 7–11 pppppppppppppppppppppppppppp985

Errata

An online log of corrections to Annual Review of Economics articles may be found at http://www.annualreviews.org/errata/economics Annu. Rev. Econ. 2019.11:55-84. Downloaded from www.annualreviews.org Access provided by Northwestern University on 10/02/19. For personal use only.

Contents vii